by Duncan Lunan

Jonathan Cowie, editor of the online journal Concatenation, commissioned me to write the first version of last week’s article in the Autumn of 2022, with the option of delaying it till Spring this year, which I had to take due to other commitments.  He sent me an illustration from another publication, comparing launch costs for vehicles at the top of the US range, past, present and future, and asked me to expand it into a full comparison of all the launchers currently available.  It proved to be a much taller order than he imagined, bordering on impossible.  When he said, ‘What would have been good is: Year: Booster: US$ valued at a defined year (the same year for all boosters) per kg.’. it might have been good in theory, but it’s not attainable in practice. 

A minor point is that the USA is not on the metric system, and since they dominate the market, launch prices are normally quoted in dollars per pound.  The conversion to kilograms is easy enough, as long as it’s not made twice  (the error which doomed Mars Polar Lander), and I did it at Jonathan’s request, but the resulting table is not compatible with most other published lists.  For the Ariane entries, the situation was the opposite:  payload masses  (or ‘throw weights’, as they’re known in the business)  were already in kilograms, but the prices were quoted in euros. 

Also, in many cases up-to-date data is simply not available.  Interestingly, quite a few of the websites had been updated since the end of February 2023, where previously they were 3-7 years out of date.  I suspect that was because the whole field of launch provision was reacting to the presence of dynamic new kids on the block.  But even then, the prices quoted came mostly from Wikipedia and many of them were dated, often quoted ‘as of 2020’ and in at least one case ‘as at 2016’.  There were still no costs quoted for China’s Long March V, because so far it’s not been used for any commercial payloads, nor for Orbex and Skyrora at the other end of the payload mass scale, even in reply to my email queries.  Trying to bring them all to a single year, and factoring in inflation, would be a huge undertaking.  Thinking about it, I realised that for the whole of the 1970s and beyond, prices were routinely quoted in 1970 US dollars, because it was so convenient – everything could be compared directly to Saturn V, still the largest rocket ever built and one of the most successful, which had almost achieved cost-effectiveness, the point at which the cost per launch began to go down, when averaged over the whole programme.  And then all the other interacting variables that I mentioned last week come into play.  It was no coincidence that the example table Jonathan had sent me compared only big boosters, all vertically launched, all with some combination of the same propellants, and crucially, all from the same launch site, Kennedy Space Center.

In any case, like-for-like comparison in terms of dollars per kilogram or pound over the whole range of available boosters was unworkable.  To take two extremes, the maximum payload of Richard Branson’s Launcher One was 300 kilograms.  There’s no way you can take the price of that per kilogram and relate it to the corresponding figure for the much larger Delta IV Heavy.  A payload as small as that could only go on Delta IV as a subsidiary payload, and what portion of the Delta IV cost it took up would depend on what else it was launching.  And there’s no point in working out the Launcher One cost of a 5-ton communications satellite because it couldn’t carry it.  I discovered that the cost per kilo of air-launches is actually very high, but that’s because the air-launched payloads are very small and what their owners are paying for is flexibility of launch site, choice of destination orbit, and above all how fast a launch can be provided.  In the mid-1980s, Prof. Gerard O’Neill’s Earth resources company went bankrupt after he built his satellite, because he couldn’t get a shared launch before the money ran out.

In addition, the figures in the table below represent the cost per kilogram of a launch using full payload capability, which will often not be the case, depending on what the payload is and what it’s for. 

To take a couple of extremes again, consider a big solar sail  (Fig. 1)  or a prototype solar power satellite.  Because it was so thin, the solar sail illustrated packed up really small  (Fig. 2), but one for a full-size interplanetary mission could be much larger – even when packed for launch, it could be so large that it needed the whole payload fairing of something like Atlas V.  The booster would have unused payload capability but couldn’t carry anything else because there was nowhere to put it.  Then, another payload could be a geodesic satellite like LAGEOS  (Fig. 3), as small and as heavy as possible to minimise atmospheric drag, so most of the payload fairing around it would be empty. 

Both could be launched by identical Atlas V’s, from KSC, and if they were going to the same destination orbit, the launch costs could be identical.  But the costs per kilogram of payload would be totally different.  Different again if one of them were to be launched from Vandenburg Air Force Base on the west coast, and still more different if one was for polar orbit and the other for Sun-synchronous, at high inclination to the equator, or into equatorial orbit, for which neither site is ideal.  And if they needed different upper stages, say one with restart capability and one without, the costs would be different again, and the difference would not be linear.  In other words, there is no such thing as a standard payload or a standard launch, and trying to reduce all the different ones to a single common denominator of dollars per kilogram or pound would be seriously misleading.

As I pointed out last week, the latitude of the launch site makes a big difference to costs.  The last Ariane Vs have now flown, launching the James Webb Space Telescope  (Fig. 4), the Jupiter Icy Moons Explorer  (Fig. 5), and finally two communications satellites into geosynchronous orbit  (Fig. 6 – see below). 

The cumulative record of all five Ariane versions since 1979 is extremely impressive  (Fig. 7). 

But one of the major factors making that possible has been the existence of the Kourou launch site, only six degrees off the equator in French Guiana.  Remarkably enough, the end of the Cold War allowed an almost like-for-like comparison in one case, because launch facilities were created at Kourou for the Soyuz-2 version of the booster which had been the workhorse of the Soviet space programme since Sputnik 1 in 1957  (Fig. 8). 

Among its successes, it launched satellites for Europe’s Galileo satellite navigation programme, and the Sentinel Earth resources satellites  (Fig. 9).  The war in Ukraine has put a lasting end to cooperation on that level.  Europe has its own small satellite launcher, Vega, which is filling the gap at the lower end of the launcher range, and the first Ariane VI, the Ariane V replacement, is already out there  (Fig. 10)  and coming up on its first prelaunch static firing.  Ariane VI will have the same payload capability as Ariane V, and be simpler, updated and cheaper – but the equatorial launch site remains the crowning advantage.

Still another big factor in comparing costs is the choice of destination orbit.  Sir Isaac Newton explained ‘orbit’ by imagining a mountain high enough to project out of the atmosphere, with a cannon mounted on the top  (Fig. 11). 

The bigger the charge of powder in the cannon, the faster the ball and the further it would go before striking the ground;  ultimately, at 5 miles per second, the curve of its path would match the curvature of the Earth and it would never come down at all, but remain in free, unpowered orbit.  (The word ‘orbit’ is not used correctly in a single episode of the original Star Trek, e.g. “The power is failing, Cap’n, we cannae maintain orbit”.)   Such an orbit is now called Low Earth Orbit  (LEO).  They can be at any inclination to the equator, and reached from any almost any launch site, but the most efficient launch from anywhere is due east, taking maximum advantage of the Earth’s rotation  (1000 mph at the equator).

At higher altitudes, in Fig. 12 ‘equatorial orbit’ is Geosynchronous Orbit, 22,000 miles up in the plane of the equator;  in the table below, GTO is Geostationary Transfer Orbit.  The payload masses quoted include the masses of the upper stages or propulsion systems which circularise the orbit at Geosynchronous distance, where the satellite remains over the same spot on the equator.  Sun-synchronous orbit   (SSO)  is at an orbital inclination and altitude such that the satellite is always in daylight, as the Earth rotates below it, and it passes over a given location at the same time each day.  It’s a special case of ‘elliptical orbit’ in Fig. 12, and also of Middle Earth Orbit  (MEO), but putting in a range of values for all the possible missions in that category would have made the table incomprehensible.

Hitherto western launches to polar, near-polar and Sun-synchronous orbit have been almost entirely from Vandenburg Air Force Base in California, to avoid overflying land after launch.  They could be launched from Woomera in Australia, over sparsely inhabited land, but the relative inaccessibility of the site has so far prevented that, except for two scientific satellites in the 1970s.  For the reasons explained last week, the UK’s coastal sites now under development are particularly suitable for these purposes. 

After Jonathan, the Concatenation editor, had grasped all that, he realised that a single graph to hold all the possible launches and costs would have to be 3-dimensional, but even if you also used different colours for different launch classes, I think you’d need more dimensions than three!  It might be produced as a chart wrapped around a globe of the Earth, like the ones of Figs. 11 and 12, but the idea reminded me of a chart of Solar System exploration to date which was produced around 2012  (Fig. 13).  The original file was gigantic and while it might have printed out as a huge wall chart, on a computer screen you could only view small parts of it at a time.  It’s packed with information, but in compressing it to be able to see it all at once, it becomes unreadable and it’s just a pretty picture.

Still, drawing up this table and the earlier versions of it wasn’t a wasted exercise.  It was the first time I’d done it since an article in 1991 for Defence and Foreign Affairs, and as well as updating, I’ve learned a lot, because there are so many more options these days.  If it’s 32 years till I’m asked to do it again, I’ll then be 110 years old, and presumably I’ll be free to decide whether to accept it as an interesting challenge, or gracefully decline because I’m too busy with other commitments.

Table of Launch Costs

Rockets, for November 5th